Sinan (@Celiksei) header
Sinan (@Celiksei) profile picture

Researching & building enterprise AI @a1consulting_

Finland
Joined December 2015
Photo by Sinan (@Celiksei)Photo by Sinan (@Celiksei)Photo by Sinan (@Celiksei)Photo by Sinan (@Celiksei)Photo by Sinan (@Celiksei)Photo by Sinan (@Celiksei)Photo by Sinan (@Celiksei)Photo by Sinan (@Celiksei)Photo by Sinan (@Celiksei)Photo by Sinan (@Celiksei)
Pinned Tweet
If you are scared of AI taking your job read this. AI boom is often compared to industrial revolutions due to massive structural changes in job market. People have a misconception of revolutions displacing workers but it's only part of the equation. Here is the full truth. 1st Industrial Revolution (1800s-1850s): Steam Power: Agriculture jobs: 70% → 40% (millions displaced) Manufacturing jobs: 10% → 32% by 1910 Net effect: Total employment grew from 5M → 15M 2nd Industrial Revolution (1900s-1950s): Electricity + Assembly Lines: Ford alone Created 500K new jobs despite displacing 200K skilled workers Net effect: Employment grew 30M → 45M 3rd Industrial Revolution (1970s-2020): Computers + Automation: Manufacturing collapsed: 30% → 9% of jobs Services exploded: 66% → 84% of employment Net effect: Employment still grew 45M → 70M The pattern is obvious. Short term losses, long term gains. Throughout every revolution: 1. Workers got displaced in dominant sectors 2. More jobs were created in emerging sectors 3. Net employment rose. In fact the highest employment rates were almost always achieved post industrial revolution: 1872, 1943, and 2018. Automation destroys job categories, not jobs. AI might take your job sure, but it will push you to adapt and do something more productive and useful for the society.
6
6
2
58
30,624
I don't care what anyone says but @OpenAI and @thsottiaux have been delivering constantly. Customer experience and communication has been top notch.
245
Fable is probably first model that is finally deviating from typical AI slop design style. Unfortunately it is expensive to use, especially just for design.
162
Everyone knows how to become successful. But it feels like a chore.
1
151
Little correction was inevitable $SPCX
1
283
What do people with extraordinarily high IQs actually do? After doing some research, I read that the average IQ among Oxford math PhDs was 128. Richard Feynman was a Putnam fellow and allegedly got a perfect score on Princeton’s grad level physics entrance exam. Yet, his measured iq was 125. Watson (co discoverer of the DNA helix) had an IQ of 124. Kasparov (world chess champion and grandmaster) at 135. This raises the question: if so many of the world’s most accomplished people fell short of “genius”, what do people who hit the mark actually do with their lives? I know many historically accomplished people probably did have IQs at this level. Still, it makes me wonder what the majority of these people do, given that it is obviously false not all of them are at the top of academia and that being measured “genius” by an IQ test is neither necessary nor sufficient for this caliber of success.
2
230
My honest read on this Anthropic situation: it ends one of two ways. They wire Trump a ton of money, or they let the government use the models in ways Anthropic explicitly prohibits. There is no third option where everyone walks away happy.
2
218
How can you tax imaginary trillions exactly?
Reason #1,000,000,000,000 why we should tax the rich.
202
Pulling Fable for everyone because foreign visa workers can't use it doesn't slow China. It doesn't slow open source. It just slow the independent devs, researchers, and small builders who were doing something real with it. The US does not realize this is a race. There's no time to pull over and have a toddler tantrum.
1
193
Fable was able to identify problems with my marketing with one prompt which Opus couldn't figure out for 3 weeks. AGI is coming baby
1
202
Three biggest IPOs happening this year will break the market. We are be witnessing massive movements of capital from established markets to these novel frontiers. Many will lose lot of money while others will win big. Exciting times ahead
177
Anyone else feel like Claude is just assuming things before even bothering with checking. The only time it actually bothers to manually verify instead of pulling from memory is when you explicitly order it to or with ultracode setting.
199
People use the dumbest AI model that exists, get a wrong answer, and declared the singularity dead. "I am so smart."
193
I think SF just a 24/7 live version of Linkedin.
1
185
This paper alone added 10$ trillion to the world's market capitalization.
1
187
Finally Claude code has proper model routing in place. No more using Opus on simple tasks.
2
187
ChatGPT just hit 1 billion monthly active users. Keep that in mind next time someone online tells you nobody wants AI.
1
171
It's funny how often people prefer AI until they find out it's AI. Even professors are getting outscored most of the time by AI with its answers being rated less harmful than the humans. It is clear that there is a bias against AI existing.
2
193
Has someone already figured out autonomous recursive agent which also handles personal responsibilities? I'm getting tired...
3
230
Bernie wants an AI sovereign wealth fund modeled on Norway's oil fund. Problem is that Norway owns its oil and he wants the government to just take 50% of private companies that nobody in government built. While some things he says are true, for letting ordinary people to decide the future of AI is not that ideal. People must be educated enough on this topic to make such decisions. They must understand in depth the benefits, the risks and the potential of AI, and we can all safely say that's not the case today.
1
286
Opus 4.8 costs less to run than 4.7 despite identical API pricing, because the CoT thinking times got streamlined. The wild part is the output quality actually went up.
1
1
305